AHA Scientific Statement

Transcription

1 AHA Scientific Statement Depression as a Risk Factor for Poor Prognosis Among Patients With Acute Coronary Syndrome: Systematic Review and Recommendations A Scientific Statement From the American Heart Association Judith H. Lichtman, PhD, MPH, Co-Chair; Erika S. Froelicher, RN, MA, MPH, PhD, FAHA, Co-Chair; James A. Blumenthal, PhD, ABPP; Robert M. Carney, PhD; Lynn V. Doering, RN, DNSc, FAHA; Nancy Frasure-Smith, PhD; Kenneth E. Freedland, PhD; Allan S. Jaffe, MD; Erica C. Leifheit-Limson, PhD; David S. Sheps, MD, MSPH, FAHA; Viola Vaccarino, MD, PhD, FAHA; Lawson Wulsin, MD; on behalf of the American Heart Association Statistics Committee of the Council on Epidemiology and Prevention and the Council on Cardiovascular and Stroke Nursing Background Although prospective studies, systematic reviews, and meta-analyses have documented an association between depression and increased morbidity and mortality in a variety of cardiac populations, depression has not yet achieved formal recognition as a risk factor for poor prognosis in patients with acute coronary syndrome by the American Heart Association and other health organizations. The purpose of this scientific statement is to review available evidence and recommend whether depression should be elevated to the status of a risk factor for patients with acute coronary syndrome. Methods and Results Writing group members were approved by the American Heart Association s Scientific Statement and Manuscript Oversight Committees. A systematic literature review on depression and adverse medical outcomes after acute coronary syndrome was conducted that included all-cause mortality, cardiac mortality, and composite outcomes for mortality and nonfatal events. The review assessed the strength, consistency, independence, and generalizability of the published studies. A total of 53 individual studies (32 reported on associations with all-cause mortality, 12 on cardiac mortality, and 22 on composite outcomes) and 4 meta-analyses met inclusion criteria. There was heterogeneity across studies in terms of the demographic composition of study samples, definition and measurement of depression, length of follow-up, and covariates included in the multivariable models. Despite limitations in some individual studies, our review identified generally consistent associations between depression and adverse outcomes. Conclusions Despite the heterogeneity of published studies included in this review, the preponderance of evidence supports the recommendation that the American Heart Association should elevate depression to the status of a risk factor for adverse medical outcomes in patients with acute coronary syndrome. Key Words: AHA Scientific Statements acute coronary syndrome coronary heart disease depression risk factors Depression and elevated depressive symptoms are common among the estimated 15.4 million US adults with coronary heart disease (CHD). 1 Approximately 20% of patients hospitalized for an acute coronary syndrome (ACS; myocardial infarction [MI] or unstable angina [UA]) meet the American Psychiatric Association s Diagnostic and The American Heart Association makes every effort to avoid any actual or potential conflicts of interest that may arise as a result of an outside relationship or a personal, professional, or business interest of a member of the writing panel. Specifically, all members of the writing group are required to complete and submit a Disclosure Questionnaire showing all such relationships that might be perceived as real or potential conflicts of interest. This statement was approved by the American Heart Association Science Advisory and Coordinating Committee on January 24, A copy of the document is available at by selecting either the By Topic link or the By Publication Date link. To purchase additional reprints, call or kelle.ramsay@wolterskluwer.com. The American Heart Association requests that this document be cited as follows: Lichtman JH, Froelicher ES, Blumenthal JA, Carney RM, Doering LV, Frasure- Smith N, Freedland KE, Jaffe AS, Leifheit-Limson EC, Sheps DS, Vaccarino V, Wulsin L; on behalf of the American Heart Association Statistics Committee of the Council on Epidemiology and Prevention and the Council on Cardiovascular and Stroke Nursing. Depression as a risk factor for poor prognosis among patients with acute coronary syndrome: systematic review and recommendations: a scientific statement from the American Heart Association. Circulation. 2014;129: Expert peer review of AHA Scientific Statements is conducted by the AHA Office of Science Operations. For more on AHA statements and guidelines development, visit and select the Policies and Development link. Permissions: Multiple copies, modification, alteration, enhancement, and/or distribution of this document are not permitted without the express permission of the American Heart Association. Instructions for obtaining permission are located at Permission-Guidelines_UCM_300404_Article.jsp. A link to the Copyright Permissions Request Form appears on the right side of the page. (Circulation. 2014;129: ) 2014 American Heart Association, Inc. Circulation is available at DOI: /CIR

2 Lichtman et al Depression Among Patients With ACS 1351 Statistical Manual of Mental Disorders (DSM) criteria for major depression, and an even larger percentage show subclinical levels of depressive symptoms. 2 4 By comparison, 4% of the general US adult population meets the criteria for major depression at any given time. 5 Numerous prospective studies, systematic reviews, and meta-analyses have shown a robust association between depression (major depression or elevated depressive symptoms) and increased morbidity and mortality after ACS. 3,4,6 11 However, depression has not yet achieved formal recognition as a risk factor for poor prognosis after ACS by national health organizations. The goal of this scientific statement is to review current evidence on the role of depression as a risk factor for adverse medical outcomes among adults recovering from ACS in order to make a recommendation as to whether depression should be elevated to the status of a risk factor among patients with ACS by the American Heart Association (AHA). Guidelines for assessing a risk factor include the presence of an objective outcome measure; prospective designs; evidence of a strong, consistent association between the risk factor and outcome; evidence that the risk factor is not explained by other variables or covariates linked to both the risk factor and outcomes; and the existence of a plausible biological mechanism to account for the observed relationship To formulate our recommendations, we examined the strength, consistency, independence, and generalizability of the findings in a set of carefully selected studies of 3 outcomes: (1) all-cause mortality, (2) cardiac mortality, and (3) composite outcomes that included mortality and nonfatal events. In addition, this Writing Group sought to identify important areas for future research that may further our understanding of the relationship between depression and ACS. Writing Group members were nominated by the committee co-chairs on the basis of their previous work in relevant topic areas and were approved by the AHA Scientific Statement Oversight Committee and the AHA Manuscript Oversight Committee. All members of the Writing Group had the opportunity to comment on and approve the final version of this document. The document subsequently underwent extensive AHA internal peer review, including Council Leadership review and Scientific Statement Oversight Committee review, before approval by the AHA Science Advisory and Coordinating Committee. Literature Search Strategy To identify relevant peer-reviewed studies, we updated previous systematic reviews 10,14 17 to include publications through July 24, We searched the MEDLINE, Current Contents, and PsycINFO databases for combinations of the following terms: 1. Coronary heart disease, coronary artery disease, ischemic heart disease, myocardial infarction, unstable angina, acute coronary syndrome, coronary bypass surgery, atherosclerosis, sudden death, ventricular fibrillation, ventricular tachycardia, or heart failure 2. Mortality, survival, or prognosis 3. Depression, depressive symptoms, dysthymia, mood, or depressive disorder The articles were limited to English language publications. The search yielded 1559 unique citations. We restricted our review to studies of patients recovering from ACS and those meeting the following additional inclusion criteria adapted from Kuper et al 17 and Frasure-Smith and Lespérance 10,14,15 : (1) They used a prospective design, (2) included 100 patients, (3) used established assessment instruments to define major depression or depressive symptoms (studies that identified depression on the basis of antidepressant treatment, self-reported treatment of depression, single-item measures, study-specific measures, or nonspecific distress screening indices were excluded), and (4) reported all-cause mortality, cardiac mortality, or a combination of either of these mortality outcomes and nonfatal events. We excluded studies that did not have a nondepressed comparison sample or that focused on the comparison of particular subtypes of depression or patterns of depressive symptoms. A minimum of 2 Writing Group members independently reviewed the titles and abstracts of each of the 1559 citations identified by the search and the 92 studies reviewed previously by Kuper et al 17 and Frasure-Smith and Lespérance 10,14,15 ; a total of 1509 citations were excluded on the basis of the predefined inclusion criteria (Figure). The full-text publications of the remaining 142 potentially eligible studies were reviewed in detail, which resulted in the exclusion of 89 additional studies. Data were abstracted from the final 53 studies by use of a standardized form that included study objective, design, data source, geographic location, and time period; sample size, definition of population/cohort, and patient characteristics; depression measurement instrument and definition, timing of the interview/ questionnaire, and depression prevalence; outcome definition, length of follow-up, and number of events; covariates included in risk-adjusted models; and results and conclusions. For each of the included 53 publications, 1 Writing Group member abstracted the data and a second reviewed the data for completeness and accuracy. Disagreements were resolved by consensus. We also conducted a systematic review to identify published meta-analyses of the relationship between depression and outcomes among patients with ACS or other CHD diagnoses. We followed the same search strategy as above, adding the following set of search terms: quantitative review or meta-analy* (using * for truncation). This search yielded 65 unique citations, including 4 meta-analyses relevant to our review. Findings A total of 53 studies met the criteria for inclusion in the present review; 32 studies reported on associations with all-cause mortality, 12 on associations with cardiac mortality, and 22 on associations with a composite outcome that included mortality and nonfatal events. These outcome groupings are not mutually exclusive. Details of the 53 studies are presented by study characteristics and by outcome type. Methodological characteristics varied across studies in terms of the selection of depression measures, outcome definitions, and covariates included in models (Table 1). The majority of studies, regardless of outcome,

3 1352 Circulation March 25, 2014 Figure. Selection of publications for abstraction. ACS indicates acute coronary syndrome; CHD, coronary heart disease. *Previous reviews included the following: Frasure-Smith and Lespérance, 10,14,15 Kuper et al, 17 and Hemingway and Marmot. 16 Outcome categories are not mutually exclusive. included 25 depressed patients in their sample, had no more than 20% of participants lost to follow-up, and adjusted for potential confounding factors. Self-report measures of depression were more common than structured or semistructured interviews, with the Beck Depression Inventory-I (BDI-I) being the most commonly used measure. Most studies assessed depression as a primary predictor of outcomes rather than merely including it as a covariate, although some of the cohorts were not specifically developed for the purpose of assessing the relationship of depression to ACS outcomes. Depression was assessed on a continuous scale in approximately half of the studies that examined all-cause and cardiac mortality outcomes; this was somewhat less common in studies that assessed a composite outcome. Few studies used troponin levels in their definition of ACS, most likely because data collection occurred before the adoption of troponin as a criterion for diagnosis. Outcome events were adjudicated independently, and adequate definitions of outcomes were provided in most analyses of allcause mortality or cardiac mortality; the quality of the outcome definitions was more varied for composite outcomes. A more detailed discussion of selected studies and results is provided below according to the specific outcome category. Depression and All-Cause Mortality The 32 studies that included all-cause mortality as an outcome (Tables 1 and 2) reported on 22 unique ACS patient cohorts. These studies included patients from 9 countries on 3 continents (North America, Europe, and Asia), with sample sizes ranging from to patients. Twenty-four studies included only patients who presented with MI,* 6 included patients with any form of ACS, 24,26,29,38,46,49 1 included only patients presenting with UA, 34 and 1 included patients presenting with MI complicated by congestive heart failure. 47 Fifteen measures of depression were evaluated as potential predictors of all-cause mortality. Self-report questionnaires were used to assess depressive symptoms in nearly all of these studies, with the BDI-I used in 20 studies. Less than half of the studies included structured or semistructured diagnostic interviews for depression. In most studies, the assessment of depression occurred during or within a few weeks after the index hospital admission. However, 3 studies measured depression before the ACS event, 18,19,47 and 2 included a measure of depression 1 year after the event. 33,35 The maximum *References 18 23, 25, 27, 28, 30 33, 35 37, 39 45, 48. References 18, 20 23, 25, 28, 29, 33, 34, 36, 39, 49.

4 Lichtman et al Depression Among Patients With ACS 1353 Table 1. Comparison of Study Characteristics by Outcomes All-Cause Mortality (n=32) Cardiac Mortality (n=12) Composite Outcome (n=22) Study sample Total sample size (38) 5 (42) 8 (36) (22) 1 (8) 8 (36) (28) 5 (42) 4 (18) (13) 1 (8) 2 (9) Sample included 25 depressed patients 30 (94) 11 (92) 19 (86) Inclusion/exclusion criteria clearly described 26 (81) 12 (100) 19 (86) No more than 20% lost to follow-up 29 (91) 11 (92) 20 (91) Definition of acute coronary syndrome Troponin levels 11 (34) 0 (0) 6 (27) Enzyme levels other than troponins 24 (75) 11 (92) 9 (41) ECG changes 25 (78) 12 (100) 12 (55) Chest pain 22 (69) 12 (100) 10 (45) Depression measure Self-report measure Beck Depression Inventory-I 20 (63) 9 (75) 11 (50) Beck Depression Inventory-II 1 (3) 0 (0) 2 (9) Center for Epidemiologic Studies Depression Scale 1 (3) 0 (0) 0 (0) Hospital Anxiety and Depression Scale, Depression Subscale 2 (6) 0 (0) 1 (5) Zung Self-Rating Depression Scale 2 (6) 2 (17) 2 (9) Other 5 (16) 0 (0) 5 (23) Structured interview Diagnostic Interview Schedule 3 (9) 3 (25) 3 (14) Depression Interview and Structured Hamilton 5 (16) 0 (0) 1 (5) Structured Clinical Interview for DSM Disorders 3 (9) 0 (0) 4 (18) Composite International Diagnostic Interview 0 (0) 0 (0) 2 (9) Other 2 (6) 0 (0) 0 (0) Depression is a primary predictor of outcomes (vs covariate) 26 (81) 11 (92) 19 (86) Depression measured by use of a continuous scale 14 (44) 7 (58) 8 (36) Outcome Outcomes were within 1 y (early recovery) 18 (56) 5 (42) 8 (36) Outcomes were assessed beyond 1 y 17 (53) 7 (58) 14 (64) At least 25 end points, with 10 events for every model variable 10 (31) 1 (8) 10 (45) Outcome events defined adequately 30 (94) 11 (92) 12 (54) Outcome events adjudicated independently (vs self-report, billing N/A 9 (75) 14 (64) codes, etc) Covariates Adjusted for potential confounding factors recognized at the time the 22 (69) 8 (67) 18 (82) study was conducted Adjusted for systolic function (eg, LVEF, Killip class) 18 (56) 7 (58) 17 (77) Risk-adjusted results Statistically significant relationship observed (depression associated with poorer outcome) 21 (65) 8 (67) 17 (77) Data are presented as n (%) and represent columnwise percentages. The number of unique cohorts represented for the 3 outcomes is as follows: 22 cohorts for all-cause mortality, 8 cohorts for cardiac mortality, and 18 cohorts for the composite outcome. DSM indicates Diagnostic and Statistical Manual of Mental Disorders; LVEF, left ventricular ejection fraction; and N/A, not applicable.

5 1354 Circulation March 25, 2014 Table 2. Summary of Included Studies That Examined All-Cause Mortality Study No. of Sites; Setting; Cohort Abrams et al, ; United States; VHA Administrative Data Berkman et al, ; United States; EPESE- New Haven, CT Enrollment Period N Mean Age, y Women, % Instrument, Prevalence ICD-9-CM /311/ 300.4: inpatient diagnosis, 5%; outpatient diagnosis, 15% , 40%; 75 84, 44%; 85, 16% Bush et al, 1; United States SCID MD/ dysthymia, 17%; BDI 10, 20% Carney et al, ; United States; ENRICHD + ancillary Carney et al, ; United States; ENRICHD + ancillary Carney et al, ; United States; ENRICHD + ancillary * 40* DISH MD, 21%; DISH md/dysthymia, 25%; BDI 10, 47% * 39* DISH MD/mD/ dysthymia, 47%; BDI 10, 47% * 40* DISH MD, 21%; DISH md/dysthymia, 25%; BDI 10, 47% Doyle et al, ; Ireland * 24 HADS-D >7/ BDI-FS >3, 18%; HADS-D >7, 15%; BDI-FS >3, 22% Drago et al, 1; Italy DSM-IV interview MD, %; BDI 10, 35% Assessment Timing In-hospital; prior 12 mo Follow-Up No. of Events Associations 1 mo; 12 mo 8100* Adj inpatient: NS (1 mo; unadj, S), NS (12 mo; unadj, S); adj outpatient: S (1 mo; unadj, NS), S (12 mo; unadj, NS ) 48 CES-D-20 16, 17% 0-3 y pre-mi 6 mo 76 Unadj: NS 2 5 d 4 mo 18 Adj BDI 10/SCID MD/dysthymia: S; adj BDI linear trend: S 28 d 30 mo 47 Adj DISH MD/ md: S; adj DISH MD: S; adj DISH md: S 28 d 30 mo (median 24) 43 Adj DISH MD/ md: NS (1 3 y); adj DISH MD/mD: NS (1 y); adj DISH MD/mD: S (2 3 y) 28 d Median 60 mo 106 Adj DISH MD/ md: S; adj DISH MD: S; adj DISH md: S 2 5 d 12 mo 24 Adj HADS-D >7/ BDI-FS >3: S; adj HADS-D >7: S; adj BDI-FS >3: NS 7 14 d Mean 60 mo 6 Adj DSM-IV MD: S; adj BDI 10: S Grace et al, 12; Canada BDI 10, 31% 2 5 d 5 y 115 Adj BDI 10: S Irvine et al, 31; Canada; CAMIAT BDI 10, NR 14 d after randomization 2 y 63 Adj BDI 10: NS Kaufmann et al, 1; United States DIS MD, 27% 3 15 d 12 mo 33 Adj DIS MD: NS (unadj: S) Kronish et al, 3; United States; BDI 10, 47%; DISH COPES MD, 11% 7 d 12 mo 18 Adj DISH MD: S; adj BDI continuous (excluding BDI 5 9): S Lane et al, 2; United BDI 10, 31% 2 15 d 12 mo 31 Unadj BDI 10: NS Kingdom Lane et al, 2; United BDI 10, 31% 2 15 d 3 y 38 Unadj BDI 10: NS Kingdom Lauzon et al, 10; Canada * 21 BDI 10, 35% 2 3 d 12 mo 28 Adj BDI 10: NS (baseline) Lespérance et al, ; Canada; EPPI DIS history of MD, 27%; DIS current MD, 16%; DIS postdischarge MD, 16% (86% of these by 6 mo); DIS current/postdischarge MD, 32% 5 15 d; 6 mo; 12 mo 18 mo 21 Unadj current MD: S; unadj 6 mo MD: NS (Continued )

6 Lichtman et al Depression Among Patients With ACS 1355 Table 2. Continued Study No. of Sites; Setting; Cohort Enrollment Period N Mean Age, y Women, % Instrument, Prevalence Assessment Timing Follow-Up No. of Events Associations Lespérance 1; Canada BDI 10, 41% Mean 5 d 12 mo 16 Unadj BDI 10: S; et al, unadj DIS MD: NS (among BDI 10) Lespérance et al, ; Canada; EPPI and M-HART BDI <5, 37%; BDI 5 9, 30%; BDI 10 18, 24%; BDI 19, 9% Parakh et al, 1; United States * 43 BDI 10, 20%; SCID MD/dysthymia/BDI 10, 27% Parashar et al, 17; United States; PREMIER In-hospital; 12 mo 5 d 12 mo; 3 y; 5 y; 8 y 5 y 155 Unadj BDI in-hospital: 5 9 vs <5: NS ; vs <5: S; 19 vs <5: S; continuous: S; unadj BDI 12 mo: 5 9 vs <5: S; vs <5: S; 19 vs <5: NS ; continuous: S 136 Adj SCID MD/ dysthymia/ BDI 10: NS (12 mo, 3 y, 5 y, and 8 y); adj BDI 10: NS (12 mo, 3 y, 5 y, and 8 y); adj BDI continuous: NS (8 y) * 33 PHQ-9 10, 22% 1 3 d 2 y 260 Adj PHQ-9 continuous, S Roest et al, 12; Canada BDI 10, 34%; 2 5 d 12 mo 51 Adj BDI continuous: S Romanelli et al, 1; United States NR BDI 10/SCID MD/ dysthymia, 23% 3 5 d 4 mo 17 Unadj BDI 10/ SCID MD/ dysthymia: S Multiple; United Rumsfeld et al, * 28* MOS-D 0.06, 23% In-hospital 2 y 98 Adj MOS-D 0.06: Kingdom, United States, Canada; EPHESUS S Shiotani et al, 25; Japan; OACIS Smolderen et al, * 20* Zung SDS 40, 42% 3 mo 12 mo 9 Unadj Zung SDS 40: NS 19; United * 32 PHQ-9 10, 22%; 1 3 d 4 y 424 Adj PHQ-9 10: S States; PREMIER Sørensen et al, 17; Denmark * 24 MDI ICD depression, 10% Steeds et al, 1; United Kingdom NR NR BDI-II 12: in-hospital, 47%; in-hospital, 2 3 mo, and 6 mo, 7% Thombs et al, 12; Canada BDI 10, 29%; BDI symptoms factor analysis, NR Median 7 d 12 mo 25 Adj MDI depression: NS (unadj: S) In-hospital ( 2 7 d); 2 3 mo; 6 mo Median 32 mo 11 Unadj BDI-II 12 in-hospital: NS; unadj BDI-II 12 in-hospital, 2 3 mo and 6 mo: NS 2 5 d 12 mo 25 Adj BDI general depression factor: S; unadj mean BDI: S Thombs et al, 12; Canada BDI 10, 34% 2 5 d 12 mo 45 Adj BDI continuous: S van Jaarsveld et al, ; Netherlands; GLAS Welin et al, 2; Sweden <55, 36%; , 64% * 38 HADS-D 8, 22% Baseline in 1993, pre-mi 1 mo; 8 y 94 Unadj HADS-D 8: NS (1 mo and 1 mo 8 y) 16 Zung SDS 40, 37% 1 mo 10 y 67 Adj Zung SDS 40: S (Continued )

7 1356 Circulation March 25, 2014 Table 2. Continued Study No. of Sites; Setting; Cohort Enrollment Period N Mean Age, y Women, % Instrument, Prevalence Assessment Timing Follow-Up No. of Events Associations Whang et al, 3; United States; * 46 BDI 10, 50%; DISH COPES MD, 14% 7 d 12 mo; 42 mo (mean 31) 12 mo, 9; 42 mo, 23 Adj BDI 10 vs 0 4: NS (12 mo; adj for only age and sex, NS ), S (42 mo); adj DISH MD: NS (12 and 42 mo) Adj indicates adjusted analysis; BDI, Beck Depression Inventory; BDI-FS, Beck Depression Inventory Fast Scale; CAMIAT, Canadian Amiodarone Myocardial Infarction Arrhythmia Trial; CES-D, Center for Epidemiologic Studies Depression Scale; COPES, Coronary Psychosocial Evaluation Studies; DIS, Diagnostic Interview Schedule; DISH, Depression Interview and Structured Hamilton; DSM-IV, Diagnostic and Statistical Manual of Mental Disorders version IV; ENRICHD, Enhancing Recovery in Coronary Heart Disease; EPESE, Established Populations for Epidemiologic Studies of the Elderly; EPHESUS, Eplerenone Post-Acute Myocardial Infarction Heart Failure Efficacy and Survival Study; EPPI, Emotions and Prognosis Post-Infarct Study; GLAS, Groningen Longitudinal Aging Study; HADS-D, Hospital Anxiety and Depression Scale-Depression; ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification; ICD-10, International Classification of Diseases, Tenth Revision; md, minor depression; MD, major depression; MDI, Major Depression Inventory; M-HART, Montreal Heart Attack Readjustment Trial; MI, myocardial infarction; MOS-D, Medical Outcomes Study-Depression; NR, data not reported; NS, statistically nonsignificant association reported; OACIS, Osaka Acute Coronary Insufficiency Study; PHQ, Patient Health Questionnaire; PREMIER, Prospective Registry Evaluating Myocardial Infarction: Events and Recovery; S, statistically significant association reported; SCID, Structured Clinical Interview for DSM Disorders; Unadj, unadjusted analysis; VHA, Veterans Health Administration; and Zung SDS, Zung Self-Rating Depression Scale. *Calculated from data provided in article. Result was marginally significant (ie, 0.05<P<0.1). Data obtained from a referenced article. follow-up period ranged from as little as 1 month 18,47 to as long as 10 years 48 after ACS. All but 10 studies adjusted for potentially confounding factors, although the covariates varied widely across studies. Of the 32 studies, 17 reported a significant risk-adjusted association and 4 reported a significant unadjusted association between at least 1 measure of depression and increased all-cause mortality. These significant findings were obtained from analyses of 13 different cohorts. Three additional studies reported a significant unadjusted association that became nonsignificant with risk adjustment. 27,28,43 Four studies reported on the results of a cohort of ACS patients in south central Ontario, Canada. 26,38,45,46 The results were largely consistent in showing a significant risk-adjusted relationship between depressive symptoms measured by the BDI-I and increased all-cause mortality, despite methodological differences across the studies, including the operationalization of depression (BDI-I 10, continuous BDI-I score, or general depression factor from confirmatory factor analysis), analytic method (survival analysis or logistic regression), inclusion of covariates, and cardiac diagnosis (entire ACS cohort or MI patients only). Hazard ratios for BDI-I scores 10 versus <10 at the time of the index hospitalization ranged from a high of 1.90 at 2 years to 1.53 at 5 years. 26 Three studies reported on a subsample of depressed patients in the Enhancing Recovery in Coronary Heart Disease (ENRICHD) clinical trial and a group of patients who were free of depression but were otherwise eligible for ENRICHD These studies found that both major and minor depression, as measured by the Depression Interview and Structured Hamilton, increased the risk of all-cause mortality over 30 months 21,22 and 5 years 23 of follow-up, even after adjustment References 19, 30, 31, 33 35, 39, 41, 44, 47. References 18, 20 26, 29, 33 35, 37 40, 42, 45, 46, 48, 49. for potential confounders and other predictors of survival. However, the significant effect of depression on mortality did not appear until nearly 12 months after the acute event. 21 Two studies each reported on patients in the Prospective Registry Evaluating Myocardial Infarction: Events and Recovery (PREMIER), 37,42 the Coronary Psychosocial Evaluation Studies (COPES), 29,49 and either the Emotions and Prognosis Post-Infarct (EPPI) cohort 33 or pooled data from EPPI and the control group of the Montreal Heart Attack Readjustment Trial (M-HART, a psychosocial intervention trial). 35 Both continuous and dichotomized scores for depressive symptoms measured during the index hospitalization by the 9-item Patient Health Questionnaire (PHQ-9) were associated with increased risk of all-cause mortality over 2 37 to 4 42 years of follow-up, respectively, in risk-adjusted analyses among >2000 MI patients in PREMIER. Among patients with ACS in COPES, both depressive symptom severity assessed by the BDI-I and major depression assessed by the Depression Interview and Structured Hamilton within 1 week of the index event were significant predictors of all-cause mortality up to 12 months after the event in analyses that adjusted for a validated cardiac risk index and systolic function. 29 When only COPES patients with UA were considered, a BDI-I score 10 (versus <5) was associated with increased risk of 42-month all-cause mortality; the association with 12-month mortality bordered on significance. 49 Major depression was not significantly associated with all-cause mortality, although the few patients reporting on major depression and the reduced sample size likely limited the statistical power of the study. Unadjusted analyses from EPPI showed a significant relationship between major depression assessed in the hospital by a modified version of the National Institute of Mental Health s Diagnostic Interview Schedule (DIS) and 18-month all-cause mortality, although depression at 6 months was not related to outcome. 33 In the larger pooled EPPI and M-HART

8 Lichtman et al Depression Among Patients With ACS 1357 sample, a dose-response relationship was observed between BDI-I scores and mortality up to 5 years after MI; depressive symptom severity at 1 year was also associated but to a lesser extent. 35 Because all-cause mortality was a secondary outcome in these studies, covariate-adjusted analyses were not presented. Although not part of the EPPI or M-HART studies, the same research team noted a similar unadjusted association of elevated BDI-I scores with mortality at 12 months among patients with UA. 34 Three studies were based on a sample of <300 patients with MI from 1 US site. 20,36,39 In risk-adjusted analyses, depression at the index hospitalization (BDI-I 10 or diagnosis of major depression, dysthymia, or bipolar disorder according to the Structured Clinical Interview for DSM Disorders) was associated with increased risk of all-cause mortality 4 months later among patients 65 years old. 20 A significant linear trend was observed for BDI-I scores, but depressive disorder diagnoses based on the Structured Clinical Interview for DSM Disorders were not associated with mortality. No association between depression and long-term mortality after MI was observed in this cohort, regardless of how depression was defined, length of follow-up (1, 3, 5, or 8 years), or covariate adjustment. 36 The remaining independent studies showing a significant covariate-adjusted association between depression and all-cause mortality included data from an administrative database of the Veterans Health Administration, 18 the Eplerenone Post-Acute MI Heart Failure Efficacy and Survival Trial (EPHESUS), 40 a multisite ACS cohort in Ireland, 24 and MI cohorts in Italy 25 and Sweden. 48 Independent studies that did not find a relationship were derived from the New Haven, CT, Established Populations for Epidemiologic Studies of the Elderly (EPESE), 19 the placebo-controlled Canadian Amiodarone Myocardial Infarction Arrhythmia Trial (CAMIAT), 27 the Osaka Acute Coronary Insufficiency Study (OACIS), 41 the Groningen Longitudinal Aging Study (GLAS), 47 multisite MI cohorts in Canada 32 and Denmark, 43 and single sites in the United States 28 and the United Kingdom. 44 Among these negative studies, sample sizes were moderate and ranged from 100 to <800 patients. Measures of depressive symptoms differed, and only 1 study incorporated a diagnostic interview. 28 Depression was not the primary predictor in 3 studies, 19,27,47 and all-cause mortality was not a primary outcome in 2 studies. 27,47 Four studies presented analyses adjusted for potential confounders. 27,28,32,43 Depression was not associated with mortality in the multivariate models, but in 3 of the studies, there were significant univariate associations before risk adjustment. 27,28,43 In several studies, unadjusted mortality rates were markedly higher among depressed than nondepressed patients, but the differences did not reach statistical significance. 32,44 Most positive studies assessed depression during or within a few weeks of the index hospitalization; in 2 studies reporting negative results, depression was measured either before MI diagnosis 47 or up to 3 months after the event. 41 Among the negative studies, only 2 used data from the same cohort. These studies examined <300 patients with MI from the United Kingdom and reported that a BDI-I score 10 (versus <10) was not associated with all-cause mortality at either 12 months 30 or 3 years. 31 In summary, the preponderance of evidence (21 of 32 published studies, including results from 17 risk-adjusted and 4 unadjusted analyses; 13 of 22 unique cohorts) suggests that depression is a risk factor for all-cause mortality after ACS. However, findings are based on a methodologically varied group of studies, and the number of mixed or negative studies highlights the complexity of the literature. Depression and Cardiac Mortality The 12 studies that included cardiac mortality as an outcome (Tables 1 and 3) reported on 8 unique ACS patient cohorts. These studies included patients from 5 countries on 3 continents (North America, Europe, and Asia), with sample sizes ranging from ,51 to patients. Eleven studies included patients who presented at the hospital with MI, 27,30,31,35,41,48,50 54 and 1 study included patients who presented with UA. 34 Only 3 distinct measures of depression were evaluated as potential predictors of cardiac mortality. Most studies included the BDI-I as a measure of depressive symptoms 27,30,31,34,35,51 54 ; the Zung Self-Rating Depression Scale was the only other depressive symptom scale that was used. 41,48 Few studies reported on major depression as assessed by structured or semistructured interviews, and all of these studies used a version of the DIS. 34,50,51 Only 2 studies reported on depression measured by both a structured or semistructured interview and a self-report questionnaire. 34,51 Depression was most commonly measured during the index hospital admission, with only 1 study measuring depression at 1 year after ACS. 35 Follow-up ranged from 6 months 30,50 to 10 years. 48 All but 4 studies 30,31,34,41 adjusted for potentially confounding factors. Of the 12 studies, 7 reported a significant risk-adjusted association and 1 reported a significant unadjusted association between at least 1 measure of depression and increased cardiac mortality. 27,34,35,48,50 53 These significant findings were obtained from analyses of 5 cohorts. Data from 3 of these cohorts were used to assess the impact of depression over multiple lengths of follow-up, with results consistent across reports. Five studies were based on either the EPPI cohort 50,51 or on pooled data from EPPI and the control group of M-HART. 35,52,53 The EPPI study used a modified version of the DIS to assess depression in 222 post-mi patients. In this key study, there were 5 times as many cardiac deaths in the depressed as in the nondepressed group over the first 6 months after an acute MI (17% of the 35 depressed patients died versus 3% of the 187 nondepressed patients). 50 Elevated depressive symptoms, as measured by the BDI-I, were also significantly associated with 18-month cardiac mortality in risk-adjusted analyses. 51 Three subsequent studies based on 896 patients with MI in the pooled sample reported consistent risk-adjusted findings. 35,52,53 Major depression almost doubled the risk of cardiac mortality over 5 years, and there was a dose-response relationship between BDI-I scores and cardiac mortality risk. A sixth study by these investigators was based on a separate cohort and found that elevated depressive symptoms nearly tripled the risk of cardiac mortality in 430 patients with UA during the year after hospitalization in unadjusted analyses. 34 The fact that these studies were conducted by the same research team and were limited to References 27, 30, 31, 34, 35, 41, 48,

9 1358 Circulation March 25, 2014 patients from the same geographic location in Canada potentially limits the generalizability of these findings. Both of the remaining positive studies assessed depressive symptoms. Depression, as measured by the BDI-I, was a significant risk-adjusted predictor of sudden cardiac death in the placebo arm but not in the active drug arm of CAMIAT. 27 A Swedish MI registry of 275 patients aged <65 years who had no prior history of MI showed a significant risk-adjusted relationship between depression measured by the Zung Self-Rating Depression Scale and cardiac mortality; however, the study relied on death certificates for the determination of cause of death, without confirmation by an independent review. 48 The 4 studies reporting negative results represent 3 distinct cohorts. In 2 studies based on a cohort of MI patients in the United Kingdom, depression did not predict cardiac mortality over any of the follow-up intervals that were studied (6 months, 12 months, or 3 years) in unadjusted analyses. 30,31 Although the studies were well designed and clearly reported, the small sample size (n=288) may have been a limitation. Despite a 4-fold difference in cardiac mortality between depressed and nondepressed patients in the OACIS cohort, the unadjusted effect of depression was not significant. 41 However, only 0.5% of the patients died of cardiac causes, and thus, there were too few events to model the effects of depression with adequate statistical power. In the final negative study, nearly two thirds of a cohort of patients with MI in Iran were classified as depressed (BDI-I 10). 54 This is approximately twice as high as the prevalence of depression that has been reported in most studies of patients after MI. One third of the total sample and a disproportionate number of the depressed patients were lost to follow-up. Furthermore, cardiac mortality was not well defined, and the deaths were not adjudicated. In summary, although few studies have systematically investigated the relationship between depression and cardiac mortality, and although results have been mixed, the preponderance of evidence (8 of 12 published studies, including results from 7 risk-adjusted analyses and 1 unadjusted analysis; 5 of 8 unique cohorts) suggests that depression is a risk factor for cardiac mortality after ACS. Depression and a Composite of Mortality and Nonfatal Events The 22 studies 25,34,35,40,41,52,55 70 that included a composite outcome of mortality and nonfatal events (Tables 1 and 4) reported on 18 unique ACS patient cohorts. Eighteen studies examined a composite end point of cardiac mortality and rehospitalization for a cardiac diagnosis, 34,35,40,41,52,55 67 and 4 examined a composite end point of all-cause mortality and cardiac rehospitalization. 25,68 70 These studies included patients from 11 countries on 4 continents (North America, Europe, Asia, and Australia), with sample sizes ranging from to patients. Thirteen studies included only patients who presented at the hospital with MI, 7 included patients with any form of ACS, 56,57,63,65,66,68,70 and 1 included only patients presenting with UA. 34 Fifteen measures of depression were evaluated as potential predictors of a composite cardiac outcome after ACS. Almost half of the 22 studies used structured References 25, 35, 40, 41, 52, 55, 58 62, 64, 67, 69. or semistructured interviews to obtain a clinical diagnosis of major depression; the DIS 34,52,55 and the Structured Clinical Interview for DSM Disorders 25,56,57,61 were the 2 most common interview schedules. Nearly all studies used self-report depressive symptom questionnaires, of which the BDI-I was the most commonly used instrument.# Only 8 studies reported on both structured or semistructured interviews and self-report questionnaires. 25,34,52,55 57,61,68 In most studies, the assessment of depression occurred during or within a few weeks of the index hospital admission, but it occurred up to 1 year after ACS in 2 studies. 35,61 The maximum follow-up period ranged from 1 year 34,41,52,55,64,65,68,69 to 5 years. 25,35 All but 4 studies 35,52,63,64 adjusted for potentially confounding factors. Of the 22 studies, 15 reported a significant risk-adjusted association and 2 reported a significant unadjusted association between at least 1 measure of depression and increased mortality or nonfatal cardiac events.** These findings were obtained from analyses of 14 different cohorts. Three studies were based on either the EPPI cohort or on pooled data from EPPI and the control group of M-HART. 35,52,55 Depression was related to cardiac events in an unadjusted analysis of the 222 patients in EPPI, with an odds ratio >2, but the relationship was not statistically significant in a multivariable analysis despite an odds ratio that remained around However, the final model included both a depression diagnosis according to the DIS and depressive symptoms measured by the BDI-I, as well as a measure of anxiety. In 2 subsequent studies of the pooled cohorts, significant unadjusted associations were observed between depression and depressive symptom severity and a composite of cardiac death and recurrent nonfatal MI. 35,52 Because both of these publications focused on cardiac mortality, no adjusted results were reported for the composite end point. Two studies each reported on the Epidemiological Study of Acute Coronary Syndromes and the Pathophysiology of Emotions (ESCAPE), 56,57 OACIS, 41,58 and a sample of MI patients in the Netherlands. 60,67 Studies on ESCAPE reported significant associations between major depression or depressive symptoms assessed 2 months after ACS and cardiac outcomes at 2 years. 56,57 Two studies from the OACIS cohort also reported consistent positive findings for depression in risk-adjusted analyses. 41,58 Although 2 studies on a sample of 473 MI patients from hospitals in the Netherlands reported results for depression, 60,67 only 1 of them 60 focused primarily on depression as a risk factor for poor outcomes. That study reported significant adjusted associations with cardiac death or recurrent MI for both the BDI-I and a shorter version, the BDI The second study focused on Type D personality but also considered depression measured with the Hamilton Rating Scale for Depression as a covariable. 67 Although depression measured with the Hamilton Rating Scale for Depression was associated with clinical outcomes in a univariate analysis, the relationship was no longer significant in multivariable models that adjusted for Type D personality in addition to cardiac risk factors. Because Type D personality #References 25, 34, 35, 52, 55, 60, 61, 63, 64, 68, 69. **References 25, 34, 35, 40, 41, 52, 56 60, 62, 65, 66,

10 Lichtman et al Depression Among Patients With ACS 1359 Table 3. Summary of Included Studies That Examined Cardiac Mortality Study Frasure-Smith et al, Frasure-Smith et al, Frasure-Smith et al, Frasure- Smith and Lespérance, No. of Sites; Setting; Cohort 1; Canada; EPPI 1; Canada; EPPI 10; Canada; EPPI and M-HART 10; Canada; EPPI and M-HART Enrollment Period N Mean Age, y Women, % Instrument, Prevalence Assessment Timing Follow-Up No. of Events Associations DIS MD, 16% 5 15 d 6 mo 12 Adj DIS MD: S * 22 DIS MD, 16%; BDI 10, 31% 5 15 d 18 mo 19 Adj DIS MD: NS (unadj S); adj BDI 10: S; adj BDI continuous: S * 32 BDI 10, 32% In-hospital 12 mo 37 Adj BDI 10: S BDI 10, 32% In-hospital 5 y 121 Adj BDI continuous: S Hosseini et al, Multiple; Iran BDI 10, 66% 15 d 2 y 55 Adj BDI 10: NS Irvine et al, 31; Canada; CAMIAT Lane et al, 2; United Kingdom Lane et al, ; United Kingdom Lespérance et al, Lespérance et al, BDI 10, NR 14 d after randomization 2 y 50 (34 SCD) Adj BDI 10: S (SCD) BDI 10, 31% 2 15 d 6 mo; 12 mo 27 Unadj BDI 10: NS (6 mo and 12 mo) BDI 10, 31% 2 15 d 3 y 33 Unadj BDI 10: NS 1; Canada BDI 10, 41% Mean 5 d 12 mo 13 Unadj BDI 10: S; unadj DIS MD: NS (among BDI 10) 10; Canada; EPPI and M-HART Shiotani et al, ; Japan; OACIS BDI <5, 37%; BDI 5 9, 30%; BDI 10 18, 24%; BDI 19, 9% * 20* Zung SDS 40, 42% Welin et al, 2; Sweden <55, 36%; , 64% 16 Zung SDS 40, 37% In-hospital and 12 mo 5 y 121 Adj BDI in-hospital: 5 9 vs <5: NS ; vs <5: S; 19 vs <5: S; continuous: S; unadj BDI 12 mo: 5 9 vs <5: NS; vs <5: S; 19 vs <5: S; continuous: S; unadj BDI residual change score: NS (overall; BDI 10 18: S) 3 mo 12 mo 5 Unadj Zung SDS 40: NS 1 mo 10 y 41 Adj Zung SDS 40: S Adj indicates adjusted analysis; BDI, Beck Depression Inventory; CAMIAT, Canadian Amiodarone Myocardial Infarction Arrhythmia Trial; DIS, Diagnostic Interview Schedule; EPPI, Emotions and Prognosis Post-Infarct Study; MD, major depression; M-HART, Montreal Heart Attack Readjustment Trial; OACIS, Osaka Acute Coronary Insufficiency Study; NR, data not reported; NS, statistically nonsignificant association reported; S, statistically significant association reported; SCD, sudden cardiac death; Unadj, unadjusted analysis; and Zung SDS, Zung Self-Rating Depression Scale. *Calculated from data provided in article. Data obtained from a referenced article. Result was marginally significant (ie, 0.05<P<0.1).

11 1360 Circulation March 25, 2014 Table 4. Summary of Included Studies That Examined a Composite Outcome That Included All-Cause or Cardiac Mortality and Nonfatal Events Study No. of Sites; Setting; Cohort Ahern et al, ; United States; CAPS Davidson et al, ; United States; COPES Enrollment Period N Mean Age, y Women, % Instrument, Prevalence * (n=502) 17* (n=502) BDI mean score BDI 10, 47%; DISH MD, 11%; DISH depressed mood, 24% Denollet et al, ; Netherlands BDI, mean score 6.5; BDI-10, mean score 2.7 Assessment Timing Follow-Up No. of Events Associations 6 60 d 12 mo 27 Adj BDI continuous: S 7 d 12 mo (mean 10.4) 67 Adj BDI 10 vs <5: NS (adj only for age: S); adj DISH MD: S (adj for anhedonia: NS); adj DISH depressed mood: NS ; adj BDI depressed mood: NS 2 mo Mean 2.7 y 41 Adj BDI: S; adj BDI-10: S Dias et al, 1; Portugal BDI 19, 15% In-hospital Mean 16 mo NR Unadj BDI 19: NS Doyle et al, 12; Ireland HADS-D >7, %; BDI-FS >3, 24% Drago et al, ; Italy DSM-IV interview MD, 15%; BDI 10, 35% Frasure-Smith et al, Frasure-Smith et al, Frasure-Smith et al, Frasure- Smith and Lespérance, ; Canada; EPPI 10; Canada; EPPI and M-HART 2; Canada; ESCAPE 2; Canada; ESCAPE DIS current MD, 15% ; DIS MD history, 27% ; BDI 10, 31% In-hospital Median 67wk 59 Adj HADS-D >7: S; adj BDI-FS >3: NS 7 14 d Mean 60 mo 30 Adj DSM-IV MD: S; adj BDI 10: S 5 15 d 12 mo 48 Adj DIS current MD: NS (unadj: S); adj BDI 10: NS (unadj: S); adj DIS MD history: NS (unadj: S) BDI 10, 32% In-hospital 12 mo 85 Unadj BDI 10: S BDI-II 14, 27%; SCID MD, 6% BDI-II 14, 27%; SCID MD, 7% 2 mo 2 y 102 Adj BDI-II 14, S (men only; unadj: S [overall and for men only]); adj BDI-II continuous: NS (men only; unadj: S [overall and for men only]); unadj SCID MD: S (men only) 2 mo 2 y 115 Adj SCID MD: S; adj BDI-II 14: S; adj BDI-II continuous: NS (unadj: S); adj SCID MD vs BDI <14 and without MD: S; adj BDI-II 14 and without MD vs BDI <14 and without MD: NS (unadj: S) (Continued )

12 Lichtman et al Depression Among Patients With ACS 1361 Table 4. Continued Study No. of Sites; Setting; Cohort Enrollment Period N Mean Age, y Women, % Instrument, Prevalence Assessment Timing Follow-Up No. of Events Associations Horsten et al, 10; Sweden; FemCorRisk Study Lane et al, ; United Kingdom Item questionnaire from Pearlin et al 71 2 symptoms, 72% 3 6 mo Median 4.8 y 81 Adj depression symptoms 2: S BDI 10, 30% 2 15 d 12 mo 82 Unadj BDI continuous: NS; unadj BDI 10: NS Leroy et al, 1; France NR HADS-D, mean score d 3 y 176 (Clinical events; includes 44 severe cardiac events) Severe cardiac events: adj HADS-D continuous: NS (unadj: S); clinical events: adj HADS-D continuous: S (NS when continuous vs categorical anhedonia adjustment) Lespérance 1; Canada BDI 10, 41% Mean 5 d 12 mo 28 Adj BDI 10: S; et al, unadj DIS MD: NS (among BDI 10) Lespérance et al, ; Canada; EPPI and M-HART BDI <5, 37%; BDI 5 9, 30%; BDI 10 18, 24%; BDI 19, 9% Martens et al, 4; Netherlands HAM-D 17, % Nakatani et al, ; Japan; OACIS (n=2509) 23 (n=2509) Zung SDS 40, 48% Parker et al, ; Australia * CIDI MD/ dysthymia: lifetime, 38%; pre-acs onset, 12%; post- ACS onset, 10% Rumsfeld et al, Multiple; United Kingdom, United States, Canada; EPHESUS Shiotani et al, ; Japan; OACIS MOS-D 0.06, 23% Zung SDS 40, 42% In-hospital; 1 y 5 y 202 Unadj BDI in-hospital: 5 9 vs <5: NS ; vs <5: S; 19 vs <5: S; continuous: S; unadj BDI 1 y: 5 9 vs <5: NS; vs <5: S; 19 vs <5: NS ; continuous: S 7 d Mean 1.8 y 44 Adj HAM-D 17: NS (unadj: NS ); adj HAM-D continuous: NS (unadj: S) 3 mo Mean 736 d 594 Adj Zung SDS 40, S Mean 4 d; 1 mo 12 mo 86 Adj lifetime: NS; adj pre- ACS onset: NS; adj post-acs onset: S In-hospital Mean 16 mo 198 Adj MOS-D 0.06: S 3 mo 12 mo 283 Adj Zung SDS 40: S (Continued )